The Quantum revolution: Live at Founders Forum

This is an audio transcript of the Tech Tonic podcast: ‘The Quantum revolution — Live at Founders Forum’

[MUSIC PLAYING] 

John Thornhill
Hello. It’s John Thornhill, the regular host of Tech Tonic here with a special bonus edition of the podcast. Earlier this year, my colleague, Madhumita Murgia and I made a series about quantum computers, asking if this revolutionary technology is really going to change the world as a lot of people claim. And a few weeks ago we hosted a session at Founders Forum, a conference for tech start-ups and investors, where we spoke to some of the people building and investing in quantum computers. We asked them about the promise of quantum computing and the state of the industry today. So this is the recording of that session. It features Steve Brierley, the boss of Riverlane, a company that’s building the algorithms and software for quantum computers; Ilana Wisby, the CEO of Oxford Quantum Circuits, a company building commercially available quantum computers; and Hermann Hauser, an investor in quantum technology. It’s a really fascinating discussion, and if you like it and haven’t yet heard our series on quantum computers and technology, all six episodes are available right now in the Tech Tonic feed. So here we go, Tech Tonic at Founders Forum with me and Madhumita Murgia.

Madhumita Murgia
Hello, everybody, and welcome to FT’s Tech Tonic podcast. We’re live from Founders Forum in sunny Oxfordshire today. And I’ve got with me three of the most fascinating people in quantum who are going to talk to us. So I’ve got right here next to me Steve Brierley, CEO of Riverlane. We’ve got Ilana Wisby, who’s the CEO of Oxford Quantum Circuits, and we’ve got Hermann Hauser, who’s the co-founder of the VC firm Amadeus Capital among other things. I’m Madhumita Murgia, the AI editor at the Financial Times.

John Thornhill
I’m John Thornhill, the innovation editor. Madhu and I have just concluded a six-part podcast series on Tech Tonic about quantum computing. And so we thought we’d pick out some of the themes that we were discussing in that series and fire them at our guest today. So, Steve, I’m going to start with you. One of the things that we discovered in going around and talking to people about quantum computing was that there’s a huge amount of excitement in this field. But the dirty secret of the industry at the moment is that they don’t actually work really very well at the moment. And so with the computers maybe have kind of 430 qubits or so on and everyone was telling us that you need a million qubits to do something really significant. Ilana is shaking her head already. Can you tell us how quickly do you think we will get to a robust quantum computer and what are the challenges that we face in getting that?

Steve Brierley
Yeah, I think it’s worth starting by saying that building a large, reliable quantum computer is one of the biggest challenges that humanity has taken on. The ability to harness the properties of nature, of atoms, transforms how we can understand nature itself, transforms how new products could be built in industries like in healthcare or in addressing climate change. Quantum computers will be used as a tool to design on a computer when previously and currently we, the best methods are to experiment in a laboratory. So, you know, when I founded Riverlane seven years ago, many people in the field thought this was impossible. A large-scale, reliable quantum computer could never be built. At the time, I thought they were wrong. And the reason is that I think what Moore’s Law teaches us is that if you improve every 12, 18 months consistently, year on year, this compounds in a way that makes the thing you build up for a decade look like magic, right? The phones we have today, it appears to be magical. And the same is happening in quantum computing. So as you said, like the best quantum computers in the world have hundreds of qubits. And indeed the number of operations we can do before failure is on the order of 100 quantum operations or quarks. And to do something really impactful, we need to get to a trillion quarks. So I think there are two major challenges to do that. The first is to increase scale. So we need bigger systems. We need more qubits. So to go from hundreds to hundreds of thousands and then ultimately millions. And the second is to improve the reliability of the system, the components, and then process the and correct errors as the computer is running.

John Thornhill
And that’s what you’re doing at Riverlane. Could you just briefly explain to us what, how you’re trying to tackle this problem?

Steve Brierley
Yes, that’s right. So our part at Riverlane in this problem is to process the data as it comes out of the quantum computer at a teraquop, at a trillion quantum operations. We’re talking about 100 terabytes per second. So just to give people a sense of that, that’s Netflix global streaming. So we need to be able to process this in real time so that we can correct errors as the computation is running. So today we have small demonstrations of this, and every 12 to 18 months we intend to 10x the size of the system. So essentially increase the scale of the problem that we can solve so that by the end of the decade we can get to a teraquop quantum computer.

Madhumita Murgia
Ilana, we, one of the things we were, we explored in our, in our series was, you know, why do we need these machines? Why are we putting all this money and effort and, you know, all of our scientific sort of news into it? What’s the excitement about? You know, your company has said that we’re on the brink of a, of an incredibly lucrative shift. What are the applications? How are we going to make money off of it?

Ilana Wisby
Yeah, I think, I think what Steve started touching on there about how fundamental this is and how, you know, this really is transformative. We’re thinking about being able to harness the underlying quantum mechanics that exists around us of nature and apply that to solving problems that are fundamentally intractable, right? It’s just it’s hugely transformative across pretty much every single market. Vertical outlet is like solving the problem of problem solving itself. And the customers that we have on our working live systems that are in data centres accessible by our customers today, right, they are running optimisation problems. They’re looking at small- scale molecular simulations on how they’re creating potentially new batteries, new pharmaceutical. We’re working with finance, they’re doing risk optimisation. They’re looking at creative portfolio analysis, working with automotive on how they’ll be able to better simulate. These are all live, you know, of course we do need full fault-tolerant quantum computation. That’s the path that we need to head towards. But I don’t believe that’s a million qubits. There is business impact significantly before that. It’s a commercial sector that is thriving in a market that is rapidly growing today. And I think we need to be a little careful about not getting lost in the noise of we do need to build a million qubits and we do need to reduce our errors. But there is impact already. And customers do need to make sure that they’re getting quantum ready, and the market needs to be quantum ready, which we’ll only do by engaging with the hardware, with developing the algorithms. And then I think in the next three to five years, and I think there was even the paper you’ve seen Jay Gambetta from IBM that was collaborative with other universities in the US, which is moving more towards quantum advantage, right? And that’s, that’s going to be a transition that we’re going to start to see. We do have this business impact in the next few years, and then ultimately, you know, the quantum future is seamless, fault tolerant. Every time you are running an existing, you know, your Excel, right? Every time you open your laptop, you’re interacting with a data centre, with a full stack, with every question, with all these different things. We have no idea what’s going on. And there’s a huge industry. There’s partnerships, there’s collaborators, there’s competitors, and that’s the quantum future, is that you’ve got your quantum processor, your classical processor, and you don’t even really need to know that it’s a quantum computer at the back end. But we have solved currently really challenging problems. And that’s, you know, an opportunity of a lifetime to be involved in creating this industry together. And I think that’s phenomenal.

Madhumita Murgia
And could I just follow up to, to ask, when we do get to a full tolerant million qubit dream world quantum computer, what problems can we solve that we can’t solve today? What is our, how are we going to go beyond making today’s problems . . . 

Ilana Wisby
First of all, I’d say it’s probably realistically, with the right operations schemes, we can bring that million down to 10,000, and that’s from our back end work and our projections for what hopefully will collaborate on some of our corrections schemes will look like. These types of problems they are, can you imagine a world where we’ve be able to create new drugs and take that 10-year life like pipeline down to simply taking it from the wet lab into a computational space? And you can start to think about how you could then apply different machine learning type algorithms. So there’s really a plethora, every single challenge that we currently face that’s either limited by data or limited by an understanding of the world around us, a quantum mechanical world around us. These are all the type of problems that you can solve, and that’s across from a materials discovery, really understanding energy band gaps for the first time. So you can use it to understand physics and then the impact of that physics that will apply but also from a data perspective. I mean, the Netflix problem is not a high-impact problem, but it’s one that generally people understand more and that’s how can I start to optimise my data and solve multivariable complex problems. You can apply that to any type of optimisation through automotive. And we’re thinking about, you know, 10, 15 years. We’ve got an increasingly connected world with more and more sensors, more and more data, more and more processing requirements that right now we won’t be able to solve or it’ll be very energy intensive to solve as well. And quantum also has a role to play in that in terms of reducing the energy intensity of these types of problems that you’d be able to solve as well.

John Thornhill
Hermann, I’d like to ask you about the investment case for quantum. Lots of VCs at the moment are getting wildly excited about generative AI. You are investing heavily in quantum. Can you tell us why?

Hermann Hauser
Well, quantum computing is one of those very fundamental step changes in computing. Having had the pleasure of contributing to the classical computing stack quite a few years ago, there are some analogies, you know, of course you need physical hardware, you need the silicon or whatever, and then you go up the stack with them through the error correction problem. And some of these things are analogous to the classical computing stack and some are really quite different. And error correction is a classic example. Looking with hindsight, we’ve got error correction on all our computers. Of course, memories are faulty, lots of errors in your memory, in your mobile phone. But we correct them with error correction schemes that, looking back, almost seem trivial compared with the problem that we have with error correction that Steven is working on a Riverlane. It’s a mega problem in quantum computing because the qubits are so fickle. You know, if you look at the reliability of a qubit and compare it with the reliability of a bit in classical computing, it’s really many orders of magnitudes worse, and you’ve got to fix that. But the reason why the investment case is so exciting is that when you have bits and you increase the number of bits, the performance of a classical computer scales linearly with the number of bits. A quantum computer scales exponentially with the number of qubits. So the actual compute power that you will finally reap from a quantum computer is exponentially higher than that of classical. Ilana said it will be able to solve many problems and actually many of the most interesting problems that classical computers cannot solve at all. So it will be a big market.

John Thornhill
And from an investment perspective, are you investing more in the hardware or software?

Hermann Hauser
In the whole stack, really. I’ve got a number of investments on the hardware side. Steph is here from Plutonic in Vancouver. I’ve invested in a number of software companies, and then there is one company in the world that’s actually an architecture company and that’s ParityQC, here in the front row, with Wolfgang and Magdalena that’s hoping to become the arm of the quantum computer field because Wolfgang found an architectural advantage of moving from the qubit space, which are these very fickle things, to the parity space, which is just a little bit up. But these parity bits are just this very much less fickle so that you can build in a great advantages with it, in particular the parallelisation. Stephen talked about teraquops that we need. Well, it depends a lot on how many computations, quantum computations you can do in parallel. And one of the things that I think you see solves is this parallelism problem.

John Thornhill
As you were saying earlier, you are involved in kind of building arm in the classical computing world. And can you make more comparisons between how that classical computing era kind of took off and where we are now in quantum?

Hermann Hauser
So one of the great excitements first for somebody like me who lived through the classical computing stack was we started in a world where IBM took the sand off the beaches, produced the chips, put it on circuit boards, created the operating system, invented the languages, wrote the software for it, delivered the computer, serviced it, and then at the end of the day, took it away from you again. So it was a completely vertically integrated stack. And the interesting development in classical computing was that these interfaces appeared first at the ingot, that IBM didn’t have to produce the pieces of silicon, but there were other companies that did that. Then there were, there was a slot for operating systems, and Microsoft became a very important company, and then there were software companies on top of it. The exciting thing about quantum computing is that these interfaces haven’t crystallised out yet, so it’s not clear yet where these interfaces will appear. I think everybody believes that error correction is got to be one of those interface problems where you, you would like to have an interface both below and above. And what I’ve learnt over many years of making venture investments is companies have a reason to exist. If there is a clear interface at the bottom and the clear interface at the top, so clear interface towards their suppliers and the clear interface towards their customers. And if you can define that interface, then you have a business.

Madhumita Murgia
And we talked, you talked, Ilana, earlier you touched on hybrid situation, right, where, and we also mentioned this briefly in our podcast, but we couldn’t get into it too much, where you have both the quantum and the classical computing working together in the interim period before we scale up. Maybe, Steve, you can jump in here and talk about what are the use cases, you know, now for sort of hybrid models of quantum and classical computers?

Steve Brierley
Yes, so I think of quantum computing and as being always hybrid, that we are not quantum mechanical beings, we are classical. And so we need to interact with the quantum computer at some a classical sense. So parts of the problem are always going to be better suited to the quantum computer and some parts to a classical computer. So we’re not going to use quantum computers to send email, but we will use them to model some of the really complex, hard parts of a system. So there are lots of ways that people are developing to make that work. So called embedding methods where you sort of model a system at different length scales and you put the really molecular modelling on the quantum computer and the kind of higher length scales on a traditional HPC. I think, as Hermann was saying, like the industry hasn’t yet figured out how best to integrate that. And I really like the work that Ilana is doing kind of working with HPC data centres to start making that happen already.

Madhumita Murgia
Do you want to talk to us a little bit about what you are doing?

Ilana Wisby
Yeah, definitely. So we certainly see that the future, it’s going to be hybrid. We need to be able to have CPUs, QPUs all interacting at same time. And QPUs being our new quantum processing unit, right? And these are used to be able to interchangeably interact between those types of units at super low latency so you can perform these type of hybrid algorithms. So with that in mind, OQC’s building production ready systems, right? We need to be able to test these systems in real world environments and build the infrastructure, and, ideally for a customer, that needs to be seamless. So if you’re a customer and you decide that you want to try and test one of these hybrid algorithms and you go to your chief information officer and you say, “Hey, I want to provide all of my customers private data and give it to this little quantum computing company out of a lab in Oxford”, they’re probably going to say no. But if you’re able to say, “Hey, we can simply add our QPU to our existing data infrastructure and then also be able to have super low latency between our existing classical compute modules and our quantum compute modules”, that’s a compelling business case, and this is exactly what we’re doing. So at OQC, we’ve deployed our next generation systems into the world’s leading infrastructure, and that is partnerships with Equinix. And we have our next generation system, Toshiko. All of our systems are named after pioneering women in Stem, and Toshiko is in Tokyo in a leading data centre, which has got 12,000 plus large corporate customers with huge data processing needs. And from a point in the very near future it’s going to be announced very shortly when Toshiko comes online, you’ll be able to simply add a QPU to your existing services. So that’s one route, is make sure you can have seamless access, low latency. We’ve already become a magnet customer, so we’ve seen quantum software companies but also existing users from our POCs buy racks right next to Toshiko. So you can get the super low latency, direct interconnect secure access, which is phenomenal. And the second thing that we’re doing, as Steve mentioned, is integrating with high performance compute. So the future of compute will be classical quantum data centres, but high performance compute centres as well. And we are deploying a different system and next generation again 32 qubits into Cesga, which is in Galicia in Spain, partnered with Fujitsu, which is a phenomenal example of both competition and co-operation. And they’re providing the quantum emulation piece. We’re providing the hardware so that the academics at the supercomputer centre will be able to interact and again do this high performance compute classical quantum integration there as well. So that just two examples of what we’re doing.

John Thornhill
I’m very interested in who’s going to win, as it were, the quantum race. I think we have a bit of a stacked panel here, but is it going to be more the kind of big tech companies who can afford to plough billions of dollars into this development, or is it going to be the start-ups who are see a real kind of niche advantage in a particular area and can build on that? So, Hermann, what do you think?

Hermann Hauser
I think the race is wide open, both in terms of which of the considerable number of technologies, be it superconducting, ion traps, neutral atoms, photonic and silicon-based electron spins will win. And these are all very good horses in the race. And then with a really innovative companies like Ilana’s, with university teams that clearly are on the top of their intellectual game, will be able to raise enough money to be real competitors with Google and IBM, or whether the sheer weight of money of large companies will win out. And you know, history has examples of both, though it is not knowable at the moment.

John Thornhill
Ilana or Steve, do you want to make the case for the start-ups?

Steve Brierley
Yeah, I think the . . . every major shift in technology, whether that’s the internet-created companies like Google or mobile networks that created essentially Arm and Qualcomm, like all of these companies were start-ups at one point, and you had a big shift in technology. Quantum is as big as those shifts. So I see every opportunity to create the next trillion-dollar company.

Ilana Wisby
If I can add, there won’t be one winner. We talk about this as if there’s a one horse winner, there’s not. There’s going to be multiple. If we look at the existing classical compute, you have different supercomputers, compute infrastructures. It’s a very rich and dense ecosystem. And I see that as the future for quantum as well. Obviously, I’m massively biased, and I certainly think I wouldn’t be committing my life to this if I didn’t think that first it was going to be used for incredibly good, but second, that there was a chance that we are certainly deeply competitive and we are going to win and very confident within that. But I think actually Vivek, the CTO of Fujitsu, said yesterday, you know, he thinks it’s the smaller companies that have the advantage here. Yes, you have IBM and Google with significant amounts of money, but that doesn’t necessarily mean it’s a done deal. And right now a lot of the challenges that we need to focus on, you can build large systems, right? You can say, “I want to build a thousand qubit system”. You’re just wasting money. You’re not getting any more performance out of that larger system. We need to iterate on smaller systems, optimise the performance, work on error correction. If you had, and the world’s most powerful quantum computers out there today are the 2130 qubit systems, they’re not the hundreds of qubit systems that exist. You’re adding cost, not performance.

Madhumita Murgia
So . . . go, go ahead.

Hermann Hauser
Just to add one, one argument in favour of the smaller companies and the diversity that Ilana mentioned. If you just look, look back at again, classical computing. The reason why classical computing managed to support such great variety of different companies was just the sheer size of the problem space that classical computing managed to address. And here we’ve got the most amazing expansion of problems that will be amenable to quantum computers. So this is potentially going to be a very large market, and small companies are traditionally much more agile to adopt the right solution for a particular problem than large companies. So my hope is on small companies, that’s for them. I’m backing because it’s difficult to back Google now.

Madhumita Murgia
Speaking of backing companies, you know, what’s the role of government here in terms of supporting, whether big or small, you know, supporting actually the development of the technology itself. Whoever, any of you can jump into that. And then the second part of that is kind of where does the UK sit globally? Are we in a position to provide that support and to drive forward the technology from here?

Steve Brierley
Yeah, I think the answer is simple. Governments should buy stuff, early prototypes. This is exactly why Silicon Valley is in Silicon Valley is because Nasa was buying the early prototypes of integrated circuits. So we know this works. You know, the UK is taking on the US, Germany. There are lots of big countries that are making large investments. China has a completely different model so kind of state-owned, effectively model and investing huge amounts of money in quantum computers. And so I think where the UK can win this being, is by being smart at where it spends its money. The UK has an advantage because it started earlier. So 2013 there was a £1bn investment in quantum technologies, and you’re seeing the benefits of that now. I mean, that’s why there are so many great quantum companies in the UK. But the UK has other advantages. So it’s great at photonics. It’s great at semiconductors or chip design because of companies like Arm. And so I think we can leverage that, and, for example, build the chips that power quantum computers.

Madhumita Murgia
Ilana, any thoughts on what government, governments should be doing?

Ilana Wisby
I certainly echo what Steve says here in the government as a customer, I think is a fantastic way to create that market pool. Ideally, the ecosystems will thrive if you make sure that you have got the healthy government market pull. And that’s what we need, and government can support that. So if you have government procurement alongside private investment, both through VC but also through customer investment, that’s going to drive innovation, that’s going to make sure as a delivery focus, we partner with the right people to deliver on that service. And that’s what the UK government are doing a really fantastic job of as well. So we see procurement both through invest through government vehicles with government as a customer. We have many contracts in that way, but also we see government playing a role as an investor as well. So we have funding through the Breakthrough fund, which again is a fantastic example of government investing into this type of technology through different mechanisms. And yeah, they were pioneering the early scheme of pushing with the quantum programme in 2013 out from academia, and I think they’ve been quite pioneering as a model that’s being replicated around the world as well with the pull.

Madhumita Murgia
And we’ll let Hermann have the last word here to wrap up what we need most in this moment to drive forward.

Hermann Hauser
And well, I’ve been lobbying for government procurement for 20 years, and it’s very nice to see that at least in quantum, it’s actually doing it. The £2.5bn that the UK has committed to the quantum industries is very welcome. And we need this innovative procurement not just for quantum but really across the tech sector. But I’m afraid I’ve got to end on a bit of a scary story about government spending. A recent McKinsey report has shown that the world, the world’s governments are going to spend $30bn on quantum computing and $15bn is spent by China. So out of the total world spend, half of it is spent by China. So we really have got to get our skates on.

John Thornhill
All right. We must wind it up there. I think it’s proof that this is an incredibly interesting field at the moment. It’s not all about generative AI at the moment, and you might not discover that from all the other tents in this session, but enormous thanks to Steve and Ilana and Hermann for a great discussion. Thank you.

[MUSIC PLAYING]

Guests
Thank you. (Applause)

John Thornhill
That was Tech Tonic at Founders Forum earlier this year. My thanks to our guests Steve Brierley, Ilana Wisby and Hermann Hauser. Listen to the whole quantum series if you haven’t already, as well as other fantastic Tech Tonic seasons, including the latest on the past, present and future of social media with Elaine Moore. I’ll be back in September with a two-part special on artificial intelligence and how it could help us talk to animals. Really amazing story. And then later in the year, Madhu and I will be back with more AI, a whole new series about superintelligent or godlike AI. We’re asking if Silicon Valley is really developing machines as intelligent or even more intelligent than humans and what that might mean for humanity. The AI apocalypse may be coming soon on Tech Tonic. Subscribe and you’ll get those episodes as soon as they’re published. Thanks for listening.

[MUSIC PLAYING]